Controllable Synthesis of Single-Walled Carbon Nanotube Framework

Oct 20, 2009 - Nanoindentation measurements by atomic force microscopy show that the spring constant of the SWNT framework film (22.6 ± 1.2 N/m) ...
0 downloads 0 Views 4MB Size
Download PDF
NANO LETTERS

Controllable Synthesis of Single-Walled Carbon Nanotube Framework Membranes and Capsules

2009 Vol. 9, No. 12 4279-4284

Changsik Song, Taeyun Kwon, Jae-Hee Han, Mia Shandell, and Michael S. Strano* Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 Received August 3, 2009; Revised Manuscript Received September 14, 2009

ABSTRACT Controlling the morphology of membrane components at the nanometer scale is central to many next-generation technologies in water purification, gas separation, fuel cell, and nanofiltration applications. Toward this end, we report the covalent assembly of single-walled carbon nanotubes (SWNTs) into three-dimensional framework materials with intertube pores controllable by adjusting the size of organic linker molecules. The frameworks are fashioned into multilayer membranes possessing linker spacings from 1.7 to 3.0 nm, and the resulting framework films were characterized, including transport properties. Nanoindentation measurements by atomic force microscopy show that the spring constant of the SWNT framework film (22.6 ( 1.2 N/m) increased by a factor of 2 from the control value (10.4 ( 0.1 N/m). The flux ratio comparison in a membrane-permeation experiment showed that larger spacer sizes resulted in larger pore structures. This synthetic method was equally efficient on silica microspheres, which could then be etched to create all-SWNT framework, hollow capsules approximately 5 µm in diameter. These hollow capsules are permeable to organic and inorganic reagents, allowing one to form inorganic nanoparticles, for example, that become entrapped within the capsule. The ability to encapsulate functional nanomaterials inside perm-selective SWNT cages and membranes may find applications in new adsorbents, novel catalysts, and drug delivery vehicles.

The controlled assembly of nanoparticles is challenging but may potentially create new classes of materials with unique combinations of properties.1,2 We have been interested in assembling single-walled carbon nanotubes (SWNTs) into three-dimensional (3-D) framework structures for molecular storage and transport applications. In the context of a welldefined, high-surface-area carbon material, advanced from highly disordered activated carbons, SWNTs have drawn much attention as promising sorbents, or storage material, for CO23-6 and H27-9 among other adsorbates. It has been demonstrated that SWNT is a better sorbent for CO2 than activated carbon or zeolite, for example.3 Despite SWNT’s favorable interaction with CO2 and H2, its application is hampered by the reduced surface area due to bundling together. Leonard and co-workers7 have proposed a solution to this problem by locking-in the swollen geometry of SWNT spun fibers, increasing the uptake of H2 storage. However, the potential for tuning the pore size using this method has not been shown. Furthermore, the SWNT fibers were initially swollen by oleum (20% free SO3, fuming sulfuric acid), the compatibility of which may limit the applicability of certain chemical groups. In this work, we propose and develop a generic strategy for constructing 3-D framework structures from SWNTs. * To whom correspondence should be addressed. E-mail: [email protected]. 10.1021/nl902518b CCC: $40.75 Published on Web 10/20/2009

 2009 American Chemical Society

Controlling the morphology of membrane components at the nanometer scale is central to many next-generation technologies in water purification, gas separation, fuel cell, and nanofiltration applications.10,11 Toward this end, we first aim to assemble SWNTs into framework-like materials with controllable pores by positioning linker molecules of different sizes between SWNT sidewalls (Figure 1a). Covalent links between SWNTs can significantly increase the framework’s mechanical strength. Intrinsic Young’s modulus of SWNTs is exceptionally high, but macroscopic SWNTs often form weakly binding bundles that easily slide against each other. Using atomic force microscopy (AFM), it has been found that intertube bridging increases the bending modulus of SWNT bundles up to 30-fold.12 We expect this covalent assembly could produce ultrathin but mechanically strong membranes that benefit high permeance gas separation. We can exploit myriad (covalent) bonding chemistries, and further tailoring of properties is possible by attaching molecules of interest to the remaining functional groups after assembly. Covalent attachment enables the attachment of small molecules because the reaction between multivalent nanomaterials (polymer, SWNT, etc.) and small molecules is still spontaneous and energetically favorable. This method is appropriate for our purpose of positioning small molecule spacers between SWNTs.

Figure 1. Synthesis of diamine-cross-linked SWNT frameworks. (a) Schematic representation of pore size control by linker approach through covalent layer-by-layer assembly. (b) Alternate depositions of CxP-SWNT and diamine linkers on an amine-functionalized substrate. CxP-SWNT was prepared by diazonium-assisted functionalization, and further activated by a uranium-based peptide coupling reagent (HATU). The linker molecules are EDA, BNZ, ODA, and TED.

In this work, thin films of SWNT framework multilayers with different linker molecules were synthesized by covalent attachment and subsequently characterized. The flux ratio comparison in a membrane-permeation experiment showed that the SWNT framework membrane with a larger linker molecule showed a larger pore structure. This synthetic method was equally efficient on silica microspheres, resulting in all-SWNT hollow capsules after silica etching. We were able to further extend to encapsulating gold nanoparticles with perm-selective SWNT cages, potentially a “cell-like” synthetic reaction system. The SWNT assembly started with diazonium-assisted functionalization13,14 of 4-carboxyphenyl groups on the SWNT surface,15 which enabled the covalent linkage with diamines through amidation (Figure 1b). The functionalization on SWNT was confirmed by Raman spectroscopy (Supporting Information). Upon reaction with 4-carboxyphenyldiazonium salt, the D peak (disorder mode, ∼1300 cm-1) 4280

was observed to greatly increase when compared to the G peak (graphitic mode, ∼1590 cm-1), which is the signature for increasing defects on the SWNT sidewalls mainly due to the carbons’ hybridization change from sp2 to sp3. The presence of carboxyl groups was also detected by Fourier transform infrared (FT-IR) spectroscopy (∼1720 cm-1, Supporting Information). The carboxyphenyl SWNT (CxPSWNT) was purified and redispersed in dimethylformamide (DMF) for the next steps. The SWNT framework thin films were synthesized using amine-functionalized silica surfaces as a template. Initially, glass slides or silicon wafers were utilized after 3-aminopropyltriethoxysilane (APTES) treatment. However, it should be noted that any surface with amine functionalities (for example, silica microspheres, thin polyester films, or anodized aluminum oxide after polyethyleneimine (PEI) treatment) was equally effective for SWNT deposition. Aminefunctionalized glass slides were immersed in the DMF Nano Lett., Vol. 9, No. 12, 2009

Figure 2. Characterizations for SWNT framework films. (a) UV-vis absorption of the film (EDA-linked) deposited on a glass slide. A broad absorption in the UV-vis range was gradually increased as the cycle repeated. Absorption at 800 nm is linearly proportional to the cycle number (inset). (b) Thicknesses of the SWNT framework film (EDA-linked) measured by spectroscopic ellipsometry. They correlate linearly with absorption at 800 nm. Inset picture shows SWNT framework films synthesized with various diamine linkers. (c) Typical AFM height image of a SWNT framework film (EDA-linked). Size is 2 µm × 2 µm and rms roughness is 4.1 nm. (d) TEM images of SWNT framework films (ODA-linked) with various cycle numbers. Scale bar ) 100 nm.

solution of CxP-SWNT for 30 min in the presence of diisopropylethylamine (DIEA, 5 mM), 2-(1H-7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HATU, 1 mM), and N-hydroxybenzotriazol (HOBt, 2 mM). We found that, without the coupling reagents, no significant deposition was observed (less than 10% when compared to coupling conditions). It is possible that hydrogen bonding or electrostatic interactions resulted in a slight deposition, but their contribution seems to be insignificant. After being washed with DMF, the glass slides were quickly immersed in the DMF solution of any particular diamine linker (5 mM) for an additional 30 min. This deposition concluded one cycle, and repetitive cycles continued afterward until desired SWNT framework films were obtained. The selected diamine linker molecules range from short ethylenediamine (EDA) to longer and rigid 4,4′-diaminobiphenyl (benzidine, BNZ) and 4-oxidianiline (ODA), to long but flexible 4,7,10-trioxa-1,13-tridecanediamine (TED), whose maximum end-to-end distances would be 1.7, 2.3, 2.3, and 3.0 nm, respectively, when fully stretched including carboxyphenyl moieties.16 However, it is likely that the molecules’ rigidity or flexibility would greatly affect the actual intertube distances. The build-up of SWNT framework layers was monitored by ultraviolet-visible (UV-vis) spectroscopy (Figure 2a). A broad absorption in the UV and visible range from CxPSWNT was gradually increased as the cycle repeated. The thickness measurements by spectroscopic ellipsometry reNano Lett., Vol. 9, No. 12, 2009

vealed the linear relationship between the thickness and UV-vis absorbance (Figure 2b). It appears that, after each cycle, the thickness of SWNT films increases by 3-4 nm on average, which is equal to or slightly larger than the thickness of a monolayer of SWNT and the diamine spacer. The aliphatic diamines, EDA and TED, tend to produce thicker films than the aromatic diamines, BNZ and ODA. We attribute such an effect to the higher reactivity of aliphatic amines, which give rise to a higher concentration of SWNT on the surface. Under the current conditions, it is highly likely that multiple SWNTs assemble at the same time before each SWNT is fully positioned on surface. This process would result in a somewhat interwoven structure of SWNT, not perfect monolayers, which may explain why each cycle of deposition yields a slightly larger thickness than a monolayer of SWNT and diamine. We observed an interconnected network structure of individual SWNTs in scanning electron microscopy (SEM, Supporting Information) and AFM (Figure 2c). They are randomly oriented but relatively uniform with apparently reduced bundling of SWNTs. Although SWNTs are not aligned with each other in the assembly, the majority of SWNTs appear to be separated individually, capable of maximizing the functional surface area in the framework structure, and potentially controlling the pore sizes. Rootmean-squared (rms) roughness for SWNT framework films with aliphatic diamines (∼4.1 nm) appeared to be larger than those with aromatic diamines (∼3.3 nm) in the AFM height 4281

Figure 3. (a,b) Force-distance indentation curves of a control SWNT (unfunctionalized) film (a, blue) and an SWNT framework film (b, red, ODA-linked). The black curves show the cantilever on the incompressible silicon wafer substrate. (c) Spring constants of the control and SWNT framework film derived from the force-distance curves. The sampling size for the control is 31, and 26 for the SWNT framework film. (d,e) A membrane-permeation experiment performed with two molecules, AniHCl and Rho-B, which are different in size. The schematic diagram of the experiment is shown in the top inset (d). The molecules were transported through a membrane (area ∼0.28 cm2) as a result of a concentration difference (0.5 mM) under ambient conditions in water. The concentration change of the inside chamber (5 mL) was monitored by UV-vis spectroscopy. The number of transported molecules was plotted versus time (d, blue dots for AniHCl and red dots for Rho-B), and the flux ratios (JRho-B/JAniHCl) were calculated for each SWNT framework membrane (e). The SWNT framework membranes were synthesized through 10 cycles with diamine linker molecules (ODA, TED, BNZ, and EDA). As a comparison, thin membranes from carboxyphenyl functionalized SWNT (f-SWNT) and pristine SWNT (control) were also tested.

images. Transmission electron microscopy (TEM) images showed a morphology change as more SWNTs were deposited with increasing cycle numbers (Figure 2d). Initially perforated, loosely packed SWNT networks became a densely packed, film-like structure, which is obvious at above 8 cycles (∼25 nm). We performed nanoindentation measurements by AFM (Figure 3a-c) to test the hypothesis that our covalent linkages increase the mechanical strength of the resulting thin SWNT films. In AFM, the applied force was measured as a function of Z-distance over SWNTs of a framework film (ODA-cross-linked). As a control, similar force-distance measurements were performed on an pristine SWNT film made by a filtration method.17 In the force-distance curves, the effective spring constants (keff) were obtained from the slopes in the linear regime. We considered the cantilever (kc) and SWNT (kSWNT) as two harmonic springs in series, and then the spring constant of SWNT (kSWNT) was extracted from the relation kSWNT ) kckeff (kc - keff)-1.18 We were able to confirm that the spring constant of the SWNT framework film increases to 22.6 ( 1.2 N/m (n ) 26, sampling number) from the control value 10.4 ( 0.1 N/m (n ) 31). Although more control experiments need to be performed, this 4282

preliminary result highlights the advantage of our covalent method, especially for ultrathin membrane fabrications. Pore size differences in the SWNT framework films have been demonstrated by a simple membrane-permeation experiment (Figure 3d,e).19 SWNT framework films were deposited onto porous anodized alumina discs (pore size 0.2 µm) with 10 cycles (∼35 nm), and the films were mounted on the bottom of the inner chamber (Figure 3d inset). We measured the trans-membrane diffusion rates of solute molecules of different sizes, aniline hydrochloride (AniHCl) and Rhodamine B (Rho-B), and their flux ratios were compared. In restrictive diffusion through pores under the same pressure on both sides,20,21 the transmembrane flux (J) is driven by the concentration difference of the solute as depicted in eq 1: J ) (De /L)∆c

(1)

where De is the effective diffusivity, L is the membrane thickness, and ∆c is the solute concentration difference. The effective diffusivity De is related to the bulk diffusivity D0 by Nano Lett., Vol. 9, No. 12, 2009

De ) D0(ε/τ)Kr

(2)

where τ is the porosity, τ is the tortuosity factor, and Kr is the restrictive factor that accounts for the effect of the pore diameter. The degree of reduction in diffusivity by the size of the solute molecule interacting with the membrane pores is reflected in the restrictive factor Kr. If we compare the flux ratios of small and large solute molecules, the pore size effect on intermembrane transport can be obtained because the other structural characteristics of the membrane in question (L, ε, and τ) are mutually canceled, as shown in eq 3. JRho-B /JAniHCl ) Kr_Rho-B /Kr_AniHCl

(3)

Diffusion rates from Rho-B and AniHCl through SWNT framework membranes were measured by UV-vis absorption changes in water under ambient conditions (Figure 3d). The obtained flux ratios are plotted in Figure 3e. It should be noted that the larger flux ratio corresponds to a smaller restrictive force and thus a larger pore diameter. From the results, we found that SWNT framework membranes generally exhibit a larger pore structure than the “control” SWNT membrane (pristine bucky paper). It is likely that SWNTs in the control membrane aggregate into bundles, resulting in a relatively compact structure. Furthermore, the pores from the functionalized but not cross-linked SWNT (f-SWNT) membrane appear to be even smaller than those from the SWNT framework membranes, which illustrates the capability of diamine linkages for creating and controlling membrane pore sizes. We also observed that, in the early stage of the film synthesis, where SWNTs are not completely packed, the restrictive force tends to be smaller. Thus a larger flux ratio was observed (Supporting Information). This result somewhat correlates with the evolving morphology in the TEM images (Figure 2d). The pore diameters of SWNT framework membranes roughly correlate with the lengths of diamine linkers. The large linker ODA (intertube distance ) ∼2.3 nm) yielded a larger pore SWNT framework membrane than EDA (∼1.7 nm). The very long but flexible linker TED furnished an intermediately sized pore membrane. However, the most rigid linker BNZ also produced an intermediately sized pore membrane. The membrane-permeation experiments show that we were able to control the pore sizes of SWNT framework membranes using different linker molecules. The framework synthesis method can be applied not only to planar surfaces, but also to microspheres, and virtually to any amine-functionalized surface, even in an irregular shape. We demonstrated this versatility with silica microspheres (diameter ∼5 µm, Figure 4). The silica microspheres were coated with PEI, and then subjected to the above covalent process (eight cycles with ODA). The SWNT framework thin films were deposited on the outer surface of the microspheres, which was easily observed by the naked eye. Even silica beads with irregular shapes (for example, silica for regular column chromatography) work equally well, which opens the possibility of large scale production. It Nano Lett., Vol. 9, No. 12, 2009

Figure 4. ODA-cross-linked SWNT framework (8 cycles) capsules with and without gold nanoparticle encapsulation. (a) Schematic representation of the synthesis of SWNT framework capsules. First, SWNT framework thin films were deposited on amine-treated (PEI) silica microspheres (5 µm diameter) (i). Then the silica core was removed by HF etching (ii). For nanoparticle encapsulation, gold nanoparticles were synthesized on the silica surface prior to the SWNT framework synthesis (iii). (b) TEM image of the ODAcross-linked SWNT framework capsule. (c,d) TEM images of gold nanoparticle encapsulated ODA-cross-linked SWNT framework capsules. All scale bars are 500 nm.

should be noted that each reaction in the covalent process for microspheres takes shorter times with a gentle vortexing, probably due to faster mass transfer. Hydrofluoric acid (HF) etching of the microspheres’ silica cores furnished hollow SWNT framework capsules, which was confirmed by TEM (Figure 4b). (CAUTION! HF is extremely toxic and should be handled with the upmost caution.) The integrity of the spherical shape appears to be maintained in the etching process. SWNT capsules have been reported by several research groups.22-28 However, such SWNT capsules were made out of SWNT-polymer composites, and conventional electrostatic interactions were utilized in the assembly. To the best of our knowledge, there is no report of hollow, all-SWNT framework capsules with covalent linkages, which are also potentially perm-selective. We envision that such perm-selective membrane capsules may find use in diverse areas from medicine and biotechnology (drug delivery) to synthetic chemistry (catalysis). For example, metal nanoparticles for catalysis can be sequestered inside the perm-selective capsule, meaning only a specific substrate will be transported and undergo a chemical transformation. As a proof of concept, we demonstrated the encapsulation of gold nanoparticles inside the SWNT framework capsules (Figure 4a bottom). First, gold nanoparticles were synthesized on the surface of silica microspheres using a slightly modified Turkevich method. Gold salts (HAuCl4) were adsorbed on the PEI-modified silica microspheres (diameter ∼5 µm), and the gold ions were reduced to gold nanoparticles by ascorbic acid in the presence of sodium citrate at room temperature. The SWNT framework thin films were deposited with ODA on these gold nanoparticle-decorated silica microspheres, and 4283

then the silica core was removed by HF etching. TEM images clearly show that gold nanoparticles (average size ∼30 nm) were encapsulated in the SWNT cages, and both individual and agglomerated gold nanoparticles were observed (Figure 4c,d). In conclusion, we have demonstrated the synthesis of porous SWNT framework thin films that can be manipulated into membranes, capsules, and nanoparticle cages. The covalent process enables us to control the pore size in the SWNT framework films, which was confirmed by a simple membrane-permeation experiment. The ability to encapsulate functional nanomaterials inside perm-selective SWNT cages is an important step toward new applications such as adsorbents, novel catalysis systems, and drug delivery vehicles. Acknowledgment. The authors are grateful for funding from British Petroleum through a grant to the MIT Energy Initiative. Partial funding from the Airforce Office of Scientific Research is also acknowledged. M.S.S. is also grateful for a Sloan Fellowship for supporting these efforts. Supporting Information Available: Detailed experimental procedures. Raman and FT-IR spectra of CxP-SWNT and SWNT framework films. AFM height images and traces. High-resolution SEM images of SWNT framework films on silica. Detailed AFM nanoindentation measurements. Membrane-permeation experiment of SWNT framework films with various cycle numbers. Nitrogen adsorption isotherms of SWNT framework films deposited on silica. This material is available free of charge via Internet at http://pubs.acs.org. References (1) Kotov, N. A., Ed. Nanoparticle Assemblies and Superstructures; CRC Press: Boca Raton, FL, 2006. (2) Rao, C. N. R., Muller, A., Cheetham, A. K., Eds. The Chemistry of Nanomaterials: Synthesis, Properties and Applications; Wiley-VCH: Weinheim, Germany, 2004. (3) Lu, C. Y.; Bai, H. L.; Wu, B. L.; Su, F. S.; Fen-Hwang, J. Energy Fuels 2008, 22, 3050–3056. (4) Dillon, E. P.; Crouse, C. A.; Barron, A. R. ACS Nano 2008, 2, 156– 164.

4284

(5) Huang, L. L.; Zhang, L. Z.; Shao, Q.; Lu, L. H.; Lu, X. H.; Jiang, S. Y.; Shen, W. F. J. Phys. Chem. C 2007, 111, 11912–11920. (6) Cinke, M.; Li, J.; Bauschlicher, C. W.; Ricca, A.; Meyyappan, M. Chem. Phys. Lett. 2003, 376, 761–766. (7) Leonard, A. D.; Hudson, J. L.; Fan, H.; Booker, R.; Simpson, L. J.; O’Neill, K. J.; Parilla, P. A.; Heben, M. J.; Pasquali, M.; Kittrell, C.; Tour, J. M. J. Am. Chem. Soc. 2009, 131, 723–728. (8) Weck, P. F.; Kim, E.; Balakrishnan, N.; Cheng, H. S.; Yakobson, B. I. Chem. Phys. Lett. 2007, 439, 354–359. (9) Yin, Y. F.; Mays, T.; McEnaney, B. Langmuir 2000, 16, 10521–10527. (10) Bruening, M. L.; Dotzauer, D. M.; Jain, P.; Ouyang, L.; Baker, G. L. Langmuir 2008, 24, 7663–7673. (11) Baxter, J.; Bian, Z. X.; Chen, G.; Danielson, D.; Dresselhaus, M. S.; Fedorov, A. G.; Fisher, T. S.; Jones, C. W.; Maginn, E.; Kortshagen, U.; Manthiram, A.; Nozik, A.; Rolison, D. R.; Sands, T.; Shi, L.; Sholl, D.; Wu, Y. Y. Energy EnViron. Sci. 2009, 2, 559–588. (12) Kis, A.; Csanyi, G.; Salvetat, J. P.; Lee, T. N.; Couteau, E.; Kulik, A. J.; Benoit, W.; Brugger, J.; Forro, L. Nat. Mater. 2004, 3, 153– 157. (13) Nair, N.; Kim, W. J.; Usrey, M. L.; Strano, M. S. J. Am. Chem. Soc. 2007, 129, 3946–3954. (14) Strano, M. S.; Dyke, C. A.; Usrey, M. L.; Barone, P. W.; Allen, M. J.; Shan, H. W.; Kittrell, C.; Hauge, R. H.; Tour, J. M.; Smalley, R. E. Science 2003, 301, 1519–1522. (15) Graff, R. A.; Swanson, T. M.; Strano, M. S. Chem. Mater. 2008, 20, 1824–1829. (16) The molecular lengths are based on the structures after AM1 semiempirical optimization. (17) Ham, M. H.; Kong, B. S.; Kim, W. J.; Jung, H. T.; Strano, M. S. Phys. ReV. Lett. 2009, 102. (18) Michel, J. P.; Ivanovska, I. L.; Gibbons, M. M.; Klug, W. S.; Knobler, C. M.; Wuite, G. J. L.; Schmidt, C. F. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 6184–6189. (19) Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278, 655– 658. (20) Chantong, A.; Massoth, F. E. AIChE J. 1983, 29, 725–731. (21) Smith, D. M. AIChE J. 1986, 32, 1039–1042. (22) Tang, M. X.; Qin, Y. J.; Wang, Y. Y.; Guo, Z. X. J. Phys. Chem. C 2009, 113, 1666–1671. (23) Shi, J. H.; Chen, Z. Y.; Qin, Y. J.; Guo, Z. X. J. Phys. Chem. C 2008, 112, 11617–11622. (24) Sugikawa, K.; Numata, M.; Kaneko, K.; Sada, K.; Shinkai, S. Langmuir 2008, 24, 13270–13275. (25) Salgueirino-Maceira, V.; Hoppe, C. E.; Correa-Duarte, M. A. J. Phys. Chem. B 2007, 111, 331–334. (26) Correa-Duarte, M. A.; Kosiorek, A.; Kandulski, W.; Giersig, M.; Salgueirino-Maceira, V. Small 2006, 2, 220–224. (27) Ji, L. J.; Ma, J.; Zhao, C. G.; Wei, W.; Wang, X. C.; Yang, M. S.; Lu, Y. F.; Yang, Z. Z. Chem. Commun. 2006, 1206–1208. (28) Correa-Duarte, M. A.; Kosiorek, A.; Kandulski, W.; Giersig, M.; LizMarzan, L. M. Chem. Mater. 2005, 17, 3268–3272.

NL902518B

Nano Lett., Vol. 9, No. 12, 2009